October 21, 2026
Most migration traffic drops are not algorithmic. They are self-inflicted: a redirect map that missed the pages carrying the most clicks, a staging environment that shipped with noindex still in the header, or an analytics property that stopped recording the day the new site went live. Those are preventable, and this SEO website migration checklist is organized around preventing them.
The short version: do the bulk of the work before launch, prioritize by organic value rather than URL count, and have monitoring running the hour you cut over. If you only remember one thing, remember that pre-launch preparation is where migrations are won or lost, because a bad URL map compounds the moment Google recrawls.
This covers platform replatforms, domain changes, HTTPS moves, and full redesigns. The mechanics overlap; the risk is always the same loss of the equity your old URLs earned.
What counts as a migration, and why each type breaks SEO differently
A migration is any change to the thing search engines use to identify and rank your pages: the URL, the platform rendering it, the protocol, or the structure linking it all together. Different triggers break different things.
| Migration type | What changes | Primary SEO risk |
|---|---|---|
| Domain change | Hostname | Every URL's authority has to transfer via redirects |
| Replatform | CMS or framework | URL structure, rendering, and template-level markup all shift |
| HTTP to HTTPS | Protocol | Mixed content, duplicate canonicals across protocols |
| Redesign on same URLs | Templates and content | On-page signals, internal links, Core Web Vitals |
| Site consolidation | Multiple sites to one | Many-to-one redirects, keyword cannibalization |
Name your type before you plan anything. A redesign that keeps every URL needs no redirect map but a brutal content and internal-link review. A domain change is the opposite. Most real projects are two or three of these at once, which is why they go wrong.
Before launch: the 70% that decides the outcome
The single highest-leverage task is a complete, verified redirect map, and it has to be built from your real data, not a crawl of the old site alone. A crawler only finds pages linked internally. It misses orphaned pages that still rank, old URLs with backlinks, and anything indexed but no longer linked.
Build your URL inventory from four sources and deduplicate:
- A full crawl of the old site (Screaming Frog or Sitebulb).
- Search Console > Performance, exported by page, last 16 months. These are the URLs earning clicks.
- Your server log files or analytics, for pages with traffic but no clicks from search.
- A backlink export (Ahrefs or Semrush) for URLs other sites point at.
Then map old to new, one to one wherever possible. A one-to-one 301 passes the most signal. Mapping ten old URLs to one new category page is sometimes right, but do it deliberately, not because mapping was tedious. For a mid-sized site of around 400 URLs, expect a redirect map to be a full day of work, and a 5, 000-URL ecommerce catalog a week with QA. Prioritize the mapping by the click and backlink data above: the top 20% of URLs by organic value usually carry 80% of the risk.
While the map is in progress, lock down the rest of the pre-launch list:
- Staging must be blocked from indexing, then unblocked at launch. Use HTTP auth or an
X-Robots-Tag: noindex, never arobots.txtdisallow, which does not reliably keep a page out of the index. The reverse mistake, shippingnoindexto production, is the most common cause of a total traffic collapse. Check it twice. - Recreate
robots.txtand the XML sitemap for the new URL structure. - Audit canonical tags on the new templates. Self-referencing canonicals on indexable pages, no stray canonicals pointing back at staging.
- Preserve or improve title tags, meta descriptions, H1s, and structured data at the template level. A new design often drops schema nobody noticed was there.
- Keep internal link equity intact. If the new navigation buries pages that used to be two clicks from the homepage, their rankings will follow.
- Benchmark everything now: rankings, organic sessions and clicks by page, indexed page count, and Core Web Vitals. You cannot diagnose a drop without a before.
A full technical SEO audit of the staging build before you launch is cheaper than diagnosing the live site after traffic falls. This is the stage where that work pays off.
Launch day: cut over in the right order
Launch is a sequence, not a switch. Do it in this order and the risk window stays short.
- Push the new site to production with redirects already active.
- Remove the staging index block. Confirm production pages return
200and are indexable. - Submit the new XML sitemap in Search Console. Keep the old sitemap available briefly so Google crawls the old URLs and sees the 301s.
- For a domain change, use the Search Console Change of Address tool after both properties are verified.
- Spot-check your highest-value redirects by hand. The top 50 URLs by clicks get a manual test, not a sampled one.
- Confirm analytics is firing on the new site and that the Search Console property matches the live domain and protocol.
Do not launch on a Friday. You want full engineering and SEO coverage for the 48 hours after cutover, when fixable problems are still cheap.
After launch: monitor, catch, and fix
The first 72 hours decide whether a problem stays small. Watch these, in order of how fast they surface trouble:
- Crawl errors and 404s.Search Console > Indexing > Pages, filtered to "Not found (404)". Any high-value URL here means a redirect is missing. Fix it the same day.
- Redirect chains and loops.Recrawl the live site. A chain of three or four hops leaks signal and slows crawling; collapse them to a single hop.
- Indexation.Search Console > Pages. Watch "Crawled, currently not indexed" and the total valid count. A slow, steady climb toward your old indexed total is normal and can take weeks.
- Rankings and clicks.Compare Search Console > Performance week over week against your pre-launch benchmark. Expect a short dip as Google reprocesses. A 10 to 15% wobble that recovers inside a few weeks is routine; a sustained drop in a specific template or section points at a real problem.
- Core Web Vitals.Confirm the new build still clears LCP under 2.5s, INP under 200ms, and CLS under 0.1. A redesign can regress layout stability badly.
Google has confirmed that site moves with 301 redirects pass ranking signals and that a well-executed move should see minimal, temporary fluctuation. What Google has not documented is an exact recovery timeline, so treat any vendor promising "rankings back in X days" with suspicion. In practice, a clean domain change on a mid-sized site tends to settle within four to six weeks; a messy one can drag for months.
Keep 301 redirects live permanently, or at least a year. Google has said redirects should stay in place for at least one year so the signal fully transfers. Removing them early reopens the loss you worked to prevent.
Frequently asked questions
What to do next
Name your migration type, pull your Search Console click export and backlink list today, and start the redirect map from real data rather than a crawl alone. That one artifact, built and verified before launch, prevents most of what goes wrong.
If the site is large or the stakes are high, bring in help before cutover rather than after a drop. Our SEO Migration work is built around exactly this sequence, and a broader Technical SEO review of the staging build will surface the template-level problems that hand-checking never catches.

